Handle enum values that contain meaningful non-alphanumeric characters

The GitHub API spec contains the following enum for sorting (truncated for clarity):

{
  "name": "sort",
  "description": "Sorts the results ...",
  "in": "query",
  "required": false,
  "schema": {
    "type": "string",
    "enum": [
      "reactions",
      "reactions-+1",
      "reactions--1",
      "reactions-smile"
    ]
  }
}

The enum generated for this results in duplicate variant names:

pub enum SearchIssuesAndPullRequestsSort {
    #[serde(rename = "reactions")]
    Reactions,
    #[serde(rename = "reactions-+1")]
    Reactions1,
    #[serde(rename = "reactions--1")]
    Reactions1,
    #[serde(rename = "reactions-smile")]
    ReactionsSmile,
}

I think this is fixable by updating the sanitize utility in typify to remove the -1 and +1 branches, and performing that conversion as a .replace("-1", "minus1"), etc in the catch all branch instead.

Start with the sanitize utility in typify and inspect how enum values are converted into Rust variant names. Reproduce the GitHub API sort enum case, then verify that values containing -1 and +1 generate distinct variants without changing their serde renames.